Experimental Validation of Hardware Trojan Detection and Aging Mechanisms
Abstract
This project will refine and experimentally validate the integrated circuit (IC) Trojan detection methodology being developed in an,on-going ONR-funded research project. In the on-going project, we are developing methodologies to detect hardware Trojans based on t,he physical phenomenon of short-term aging. Short-term aging effects occur as a combination of aging degradations, which are induced, at the high voltage, along with the switching to the lower voltage. Aging-induced degradation can differently influence the delay o,f gates; therefore, aging can impact IC path delays differently. We use observations of output bit errors (measured under IC stresse,s such as short-term aging and overclocking) as a probabilistic signature of the IC operation and flag deviations from expected patt,erns of bit error behaviors to detect Trojans. The on-going project has demonstrated efficacy of short-term aging based hardware Tro,jan detection via high-fidelity physics-based simulation studies on ICs on several types of hardware Trojans. The simulation studies, considered significant IC-to-IC and on-IC variations through stochastic perturbations injected into the simulation studies.The prop,osed project will demonstrate and validate our methodology for non-destructively evaluating physical ICs to detect deviations from e,xpected baseline behavior. Our methodology infers internal properties of ICs by: (a) applying excitations (e.g., overclocking, volta,ge overscaling, short-term aging, temperature variations) that induce stresses resulting in output bit errors from ICs, (b) collecti,ng input/output data to build signatures of output bit errors, (c) training machine learning (ML) models that analyze the observed p,atterns of output bit errors to thereby infer internal properties of ICs (e.g., presence of Trojans). The ICs are either procured fr,om vendors or developed as part of a concurrent DURIP project. The DURIP project designs and fabricates ICs that contain good and Tr,ojan-infected variants of cryptographic accelerators, microcontrollers and memory elements. The DURIP testbeds primary focus is to e,nable experimental validation of the hardware Trojan detection methodology. Another objective of the project is to study the role of, memory in the context of aging mechanisms and Trojans.Additionally, the project will develop and experimentally validate the SHAZAM,M (Security Health Assessment and Zooming in on Age of Modern Microelectronics) technique to apply the bit error based IC evaluation, methodology to non-destructively assess functional age of ICs. We will observe I/O bit error patterns under IC stress conditions as, an indicator/signature of age.We will work with the Naval labs to transition the technology to practice for applications integral t,o the Navy. For this purpose, the NYU team has had discussions with Navy technical staff and logistics specialists to apply our meth,odologies.The input/output based IC security/health assessment methodology can provide a crucial capability to mitigate threats such, as supply chain vulnerabilities. The tools being developed can play a vital role in a wide range of Navy applications since ICs are, widely used in all on-board systems in ships, aircrafts, and underwater vehicles.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 01, 2022
- Source ID
- N000142212314
Entities
People
- Farshad Khorrami
Organizations
- New York University
- Office of Naval Research
- United States Navy